A bilingual, traceable index of research outputs from Zhongguancun Academy (北京中关村学院) and the Zhongguancun Institute of Artificial Intelligence (中关村人工智能研究院).
Public-source coverage is maximized, but absolute completeness cannot be guaranteed. Every item retains inspectable institution evidence.
- Search structured affiliation metadata across Crossref, OpenAlex, Europe PMC, arXiv and DataCite.
- Backfill papers explicitly announced by the Academy's official research feed, then resolve their arXiv IDs and DOIs.
- Scan official project repositories for exact affiliation lines that are present in a paper PDF or README but absent from arXiv metadata.
- Merge formal publications, preprints and companion datasets without deleting existing records when a source is temporarily unavailable.
- TacReasoner: A Dynamic Tactile-Language Framework for Interactive Reasoning in Real-World Scenarios — IROS 2026 (2026)
- Feeling the Unexpected: ResTacVLA for Contact-Rich Manipulation via Residual Tactile Representation — IROS 2026 (2026)
- UrbanWell: Benchmarking Multimodal Large Language Models for Spatio-Temporal Urban Wellbeing Analytics — KDD Datasets and Benchmarks Track 2026 (2026)
- SciCustom: A Framework for Custom Evaluation of Scientific Capabilities in Large Language Models — ACL 2026 (2026)
- IntentVLA: Short-Horizon Intent Modeling for Aliased Robot Manipulation — arXiv (2026)
- PipeSD: An Efficient Cloud-Edge Collaborative Pipeline Inference Framework with Speculative Decoding — ICML 2026 (2026)
- Data Preparation for Large Language Models — Journal of Computer Science and Technology (2026)
- FutureWorld: A Live Reinforcement Learning Environment for Predictive Agents with Real-World Outcome Rewards — arXiv (2026)
| Source | Status |
|---|---|
| arxiv | ok (0 matched) |
| bza_official | ok (21 matched) |
| core | optional key |
| crossref | unavailable (HTTPError) |
| datacite | ok (17 matched) |
| europe_pmc | ok (0 matched) |
| github_projects | ok (7 matched) |
| lens | optional key |
| openalex | ok (0 matched) |
| semantic_scholar | optional key |
public/data/works.json— normalized metadatapublic/data/works.bib— BibTeX exportpublic/data/stats.json— collection statisticspublic/feed.xml— Atom feed
npm ci
python3 scripts/pipeline.py build
npm run devRun networked discovery with python3 scripts/pipeline.py fetch. Optional API keys are documented in .env.example and should be stored as GitHub Actions secrets.
Historical affiliation discovery uses the partner-university list published by
bjzgcai
as a structured prefilter. OpenAlex first selects papers involving those institutions and an arXiv location; only that reduced queue is allowed to request https://arxiv.org/html/<id>v1.
python3 scripts/arxiv_html_backfill.py prefilter --from 2024-06-01
python3 scripts/arxiv_html_backfill.py start
python3 scripts/arxiv_html_backfill.py statusThe local SQLite checkpoint is resumable and ignored by Git. arXiv requests are single-connection, at least 3.5 seconds apart by default, take a five-minute rest every 500 requests, and back off automatically after throttling or server errors. Only HTML matches in the author-affiliation region are published automatically; body-text matches remain audit-only associations.
Open an issue or pull request. Stable overrides live in data/overrides.yml and are never replaced by the automated pipeline.
Code is MIT licensed. Aggregated metadata remains subject to its original source terms and always retains provenance links.
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?v=4){kind=small}
